LightSeq2: Accelerated Training for Transformer-Based Models on GPUs.
Xiaohui WangYang WeiYing XiongGuyue HuangXian QianYufei DingMingxuan WangLei LiPublished in: SC (2022)
Keyphrases
- data sets
- training algorithm
- object detection
- general purpose
- machine learning
- prior knowledge
- training examples
- computational power
- autoregressive
- training process
- parallel processing
- statistical models
- experimental data
- statistical model
- complex systems
- power system
- database
- fuzzy logic
- hidden markov models
- active learning
- real time